FR2855001A1 - METHOD FOR RECEIVING TWO CORREL SIGNALS TRANSMITTED ON A SINGLE CHANNEL AND RECEIVER FOR CARRYING OUT SAID METHOD - Google Patents

Abstract

A method of receiving a composite signal transmitted over a nonlinear transmission channel including a first UL signal and a second LL signal. The method comprises the following steps: - demodulation and decoding of said first UL signal by means of a first demodulation and decoding chain (2, 3) so as to regenerate said first UL information; - recoding (4) and formatting to produce a continuous-time waveform; - apply to said continuous-time waveform a non-linearity function (5) based on a set of coefficients updated according to an adaptive correlation calculation method - subtract from said composite signal the result of said non-linearity function to generate a result E; - demodulation and decoding of said result E by means of a second demodulation and decoding chain (8, 9) so as to regenerate said second information LL.

Description

Method for receiving two correlated dec signals

  Technical field of the invention The present invention relates to the field of telecommunications and 1o in particular a method of reception of two uncorrelated signals transmitted on a single channel.

  STATE OF THE ART Bandwidth requirements are increasing in parallel with the development of telecommunications.

  In general, the bandwidth of a given transmission channel is increased by increasing the number of bits transmitted simultaneously on a digital transmission channel (operation which is commonly referred to as bit loading in the English literature) according to a given modulation scheme defining a finite set of complex points or symbols carrying information. In phase shift modulation (MDP or Phase Shift Keying in Anglo-Saxon literature) we define a set of complex points distributed on the same circle. If the simplest version called MDP2 (BPSK Binary Phase Shift Keying according to English literature) allows to transmit 1 bit per symbol, the MDP4 (or QPSK) transmits two 30 bits simultaneously, we can reach 3 bits with 8QPSK modulation and four bits with 16QPSK modulation.

  ST - 03-GR2-081 The modulation of two carriers in quadrature, called QAM (Quadrature Amplitude Modulation in Anglo-Saxon literature) allows to further increase the transmission rates by means of a larger constellation by operating at both modulation by phase shift and amplitude. A 16QAM or 64QAM modulation makes it possible to increase the bit rate up to 4 or 6 bits of information respectively, especially when a powerful system of error correcting code is associated with the modulation scheme.

  QAM techniques cannot however be envisaged when the transmission channel suffers from unacceptable linearity, as is encountered in certain situations, and in particular in satellite transmissions. As is known, the electronics which are on board to operate in a satellite are subject to strict operating constraints, and in particular a minimum power consumption due to the supply by solar battery. 15 Thus, the traveling wave amplifiers (or traveling wave tube TWAT according to the Anglo-Saxon name) which are on board the satellites operate in a saturated mode in order to minimize the consumption of electronic current. Non-linearity is then very important, as illustrated in FIG. 1 o, on the abscissa, there is the amplitude of an input signal 20 in the transponder. On the ordinate, the curves respectively represent the output amplitude and its statistical distribution as well as the phase curve of the transponder output signal. As can be clearly seen, around the saturation point set at 0 dB, the transponder has a saturation corresponding to a maximum amplitude of the output signal from this transponder.

  In the presence of such a nonlinearity, one cannot have recourse to the constellations and the performances of the QAM type constellations. We are then forced to limit ourselves to simpler modulation schemes, compatible with the non-linearity affecting the transmission channel, and in particular to phase shift modulation (MDP) or MPSK. In practice, for a satellite transmission, the performance of an 8PSK modulation is hardly exceeded, which corresponds to the transmission of three bits of information simultaneously.

  To increase throughput, other alternative techniques should be considered.

  ST - 03-GR2-081 This is the object of the present invention which aims to remedy the limits posed by the non-linearity of certain transmission channels which, moreover, is likely to vary over time.

  SUMMARY OF THE INVENTION The purpose of the present invention is to propose a technique making it possible to increase the bit rate in a non-linear transmission channel and to propose an alternative solution to QAM type modulations.

  Another object of the present invention consists in carrying out a method of receiving, through a single channel suffering from non-linearity, a first and second data stream transmitted according to an MPSK type modulation scheme, with a device automatic adjustment of parameters according to the non-linearity of the channel.

  The invention achieves these aims by means of a method of receiving a composite signal transmitted via a non-linear transmission channel according to a modulation scheme using a given constellation comprising a first signal UL and a second signal LL.

  The method comprises the following steps: - demodulation and decoding of said first UL signal by means of a first demodulation and decoding chain so as to regenerate said first UL information; recoding from said first regenerated UL information a set of symbols z = x + j y representative of said constellation used in transmission and formatted to reconstruct a continuous time waveform of said reconstituted symbols; - applying to said continuous time waveform a non-linearity function; ST - 03-GR2-081 - subtract from said composite signal the result of said non-linearity function to generate a result E; demodulation and decoding of said result E by means of a second demodulation and decoding chain so as to regenerate said second information s LL.

  In a preferred embodiment, the non-linearity function is performed by means of a correspondence table comprising p coefficients Cn (k) with k = 1 to p which are updated according to an adaptive process based on a calculation of 1o correlation between the result E of said subtraction and said continuous time waveform.

  In this way, it is possible to decode the two pieces of information UL and LL without, however, knowing precisely the non-linearity which affects the transmission channel. This also allows the process to operate even when the internal characteristics of the channel vary over time.

  Preferably, the coefficients of the correspondence table will be available more densely around the saturation point of the transmission channel so as to benefit from the greatest precision around this point.

  In a particular embodiment, the application of said coefficients Cn (k) of the non-linear function is carried out by means of a double Cartesian-polar-Cartesian transformation comprising the following steps: transformation of said input signal into polar coordinates to determine the modulus and phase of said input signal; - application of the amplitude response F (p) and of the phase response 0 (p) corresponding to the coefficient Cn (k) to be used; - polar / Cartesian conversion of the result.

  ST - 03-GR2-081 In a preferred embodiment, the current coefficient Cn (k) used for said non-linearity function is updated according to the following mechanism: C, + 1 (k) = Cn (k) + p E z * With t being a time constant, E being the result of said subtraction and z * being the conjugate complex of said symbol z = x + j y. 10 Alternatively, the computation time can be reduced by directly addressing the correspondence table from the value x2 + y 2 instead of that of the module of z. The method finds a direct application in the satellite transmission of a composite signal comprising a first primary information UL which can be received and processed by a large number of television receivers and secondary information LL which can be received and processed by a sub- category of receivers.

  The invention also provides a digital communication system receiver intended to receive and process a composite signal comprising a first signal UL and a second signal LL. The receiver is characterized in that it comprises: a first demodulation and decoding chain for demodulating and decoding said first UL signal so as to regenerate said first UL information; means for recoding and formatting making it possible to regenerate a continuous time waveform of said reconstituted symbols; - a non-linearity function applied to said continuous time function; - means for subtracting the result of said non-linearity function from said composite signal in order to generate a result E; ST - 03- GR2-081 - a second demodulation and decoding chain for carrying out the demodulation and decoding of said result E so as to regenerate said second information LL.

  Preferably, as for the method, the receiver comprises a set of coefficients which are updated according to a correlation calculation between the result E of said subtraction and said continuous time waveform, according to the mechanism defined below: | 0 Cn + 1 (k) = Cn (k) + p E z * With p being a time constant, E being the result of said subtraction and z * 15 being the conjugate complex of said symbol z = x + j y.

  The receiver can be used for any communication system intended for a non-linear channel, and in particular for a television receiver transmitted by satellite by non-linear traveling wave amplifiers.

Description of the drawings

  Other characteristics, objects and advantages of the invention will appear on reading the description and the drawings below, given solely by way of nonlimiting examples. In the appended drawings: FIG. 1 represents the typical curve of non-linearity of a traveling wave amplifier 30 typically used in the on-board electronics of a satellite.

  FIG. 2 illustrates the principle of an 8-state phase shift modulation allowing the simultaneous transmission of three bits of information.

  ST - 03-GR2-081 Figure 3 illustrates a circuit for receiving two signals transmitted simultaneously on a single non-linear transmission channel, using an adaptive compensation circuit.

  FIG. 4 illustrates a preferred embodiment of the block 5 for modeling the non-linearity function of the transmission channel.

  FIG. 5 illustrates the operation of the adaptive control of the coefficients Cn (k) used for the non-linearity function 5.

  FIG. 6 illustrates the method for adaptively updating the set of coefficients of the correspondence table used for the non-linearity function.

  Description of a preferred embodiment

  Referring to FIG. 3, a circuit for receiving a composite signal comprising two pieces of information transmitted simultaneously on a single transmission channel of non-linear characteristic will now be described. It will be noted that the reception circuit described below is particularly suitable for reception of a composite signal transmitted by satellite. In a given embodiment, the composite signal consists of a first signal called UL (Upper Layer) - preferably of preponderant amplitude - and a second signal 30 called LL (Lower Layer) of reduced amplitude. Typically, the signal UL is the preponderant signal of a first piece of information such as a television message intended to be received and treated by a whole category of television systems whereas the second signal LL carries a second piece of information that only will be able to decode a particular subcategory of television systems.

  ST - 03-GR2-081 In general, the first and second information UL and LL are any information which is chosen, however, so as to show no correlation.

  As can be seen in FIG. 3, the reception circuit - which may be in electronics positioned on earth - comprises a first demodulation chain making it possible to process the received composite signal (represented by the reference 1). This first demodulation chain consists of a demodulator UL 2 associated with an error correction circuit UL FEC 3. The demodulator 2 and the error correction circuit 3 are circuits known to those skilled in the art and will therefore not be discussed further. In general, as is known to a person skilled in the art, demodulator 2 is associated with a sufficiently powerful error correcting code - such as Reed 15 Solomon or even turbo-code for example, in order to be able to overcome the shortcomings of the transmission channel and in particular to correct the disturbances emanating from the second information LL seen as "noise" for the demodulation chain 23.

  The demodulation and error correction chain 2-3 therefore makes it possible to eliminate this "noise" and, consequently, to regenerate the first information UL which is available on the electrode 11 for example, and can be broadcast to any system or application receiving this UL information.

  The reception circuit of FIG. 1 further comprises a circuit for recoding the primary information UL. The latter consists of a UL 4 encoder designed to reconstruct the symbol sequence of the constellation and transform it into a continuous time function with a given spectrum, as is done in the transmitter. The shaping of a physical signal from a sequence of symbols is an operation well known to a person skilled in the art and will not be developed further. As is known to a person skilled in the art, the coding and shaping block 4 performs processing similar to that used by the transmitter according to the modulation scheme considered, for example 8PSK.

  ST - 03-GR2-081 The continuous time signal generated by block 4 is then transmitted to a block performing a non-linearity function and more specifically non-linear processing of the same kind as that which affects the transmission channel considered and, therefore, the composite signal 1.

  In general, numerous ways of modeling the non-linearity of the channel are conceivable, and in particular when the characteristic curves of the traveling wave transponders are known to the manufacturers.

  In a preferred embodiment, the non-linearity function of the transponder is carried out by means of a look-up table making it possible to associate with a value the input amplitude of a signal ( or the module of a complex Iz l = Ix + jy 1) a given complex coefficient. This results in a modeling of non-linearity by means of a successive approximation curve as shown in FIG. 4. A set of p complex coefficients Cn (k) with k = 1 to p, correspond to as many "steps "approaching the amplitude and phase curve of the transponder.

  The index n takes into account the variation of the coefficient over time by means of an adaptive process which will be described later. Preferably, the different "stages" are distributed in a non-uniform manner to provide maximum precision around the point of saturation of the amplifier used in the satellite as shown in Figure 4.

  We will now develop further the variation of the coefficient Cn (k) as a function of k, then of n.

  1) Variation of the coefficients Cn (k) with the index k 30 The coefficients Cn (k) are corrected in such a way that the complex value of the non-linearity output signal is equal to the input value multiplied by the coefficient Cn (k) corresponding to the module of the input ST - 03-GR2-081 In Figure 4, if we call F (p) the amplitude response and 0 (p) the phase response, the factor C ideally has for module not F (p), but HERE = F (p) / Ipl and for phase 0 (p), p being the module of the input; this solution is chosen because the data enters and leaves in Cartesian coordinates (real part 5 and imaginary part), we simply have to calculate a module square to choose the coefficient and perform a complex multiplication to obtain the output: X + jY = C (x + jy) Alternatively, we can also transform the coordinates into polar, then directly calculate the new module by tabulating the curve F (p), and add 0 (p) to the argument (or multiply by exp (j0 (p) )). We can then go back to Cartesian coordinates but with a higher implementation cost due to the numerous calculations implemented.

  Note that we can also work by developing at the first order: instead of storing only C we also keep a coefficient C ': we would thus have X + jY = (C + C'dr) (x + jy) where dr is the difference between the square of the module measured with respect to the square of the module of the center of class C (k) 20 The update of C is identical, and that of C 'is C'n + 1 (k) = C'n (k) + Jy 'dr E z * In this way it is possible to envisage fewer entries in the table, but 25 at the cost, however, of greater complexity.

  The reception circuit further comprises a subtractor 7 which receives at the output of block 5 the reconstructed, recoded and processed signal as well as the composite signal 30 assigned a delay by means of a delay line 6.

  The output of the subtractor 7 then provides a signal which, in theory, is free of the primary information UL and which can therefore be used by a second demodulation and decoding chain to regenerate the secondary information LL. This second ST-03-GR2-081 chain includes a UL 8 demodulator associated with an LL FEC 9 block ensuring the implementation of an error correcting code similar to the set 2-3 of the demodulation chain for the primary information.

  2) Variation of the coefficients Cn (k) with the index n In a preferred embodiment, the set of coefficients which is stored in the correspondence table is constantly updated by means of an adaptive mechanism which makes it possible not to know the values of the internal characteristics of the non-linear transponders used in satellites.

  To this end, a servo loop is inserted into the adaptive mechanism aimed at minimizing the correlation between two signals present in the reception circuit, and in particular the signal at the output of the subtractor 7 and the signal 5s at the output of the block 5 of nonlinear modeling.

  More specifically, we apply the following correction mechanism to calculate the new value Cn + 1 (k) of a coefficient used by the mechanism at a given time: Cn + 1 (k) = Cn (k) + t E z * With g being a time constant allowing a correction of the integral proportional type.

  E being the error signal at the output of subtractor 7 and z * being the conjugate complex of the signal z = x + j y reconstructed by block 4.

  If the input signal of block 5 is: z = x + j y, at output block 5 delivers a value Z = X + j Y, with Z = Cn (k). z o Cn (k) is the coefficient of the coefficient table corresponding to the value of the input module of z. ST - 03-GR2-081 - 12 In general, we note that the time constant proper to this regulation mechanism is proportional to g * | z 2. We observe a faster convergence of the mechanism if we decrease p with the modulus of z. The operation of the adaptive mechanism is illustrated more particularly in the block diagram of FIG. 5. A block 507 supplies the reconstructed signal z (after shaping by block 4) to a block 502 which performs the calculation of the module of z: Iz x = (2 + y2) withz = x + jy The value of the module is then used to select the value of k to be used for the choice of the complex coefficient Cn (k).

  A block 501 multiplies this complex coefficient 501 by the signal z so as to produce the value Z according to the formula: Z = Cn (k) xz This value is then subtracted from the composite signal supplied by block 20 508 by means a subtractor 504 which then supplies an error signal E which can be transmitted to the decoder LL.

  The same signal E is also transmitted to a multiplier 505 which receives the conjugate complex of the signal z so as to perform a correlation calculation Ex Z * The result is transmitted to an attenuator block 509 to carry out the multiplication by the coefficient, t and the result is then added to the current value 30 Cn (k) of the coefficient which was used in the servo loop. Thus, an adaptive control of the value of the current coefficient Cn (k) is carried out so as to calculate the new value Cn + 1 (k) which must then be loaded into the table.

  ST - 03-GR2-081 - 13 Figure 6 is a representation of the adaptation mechanism following a formalized approach following a process. At the start, the coefficients Cn (k) are fixed at a predefined value, possibly possibly approaching the internal average characteristics of the transponders used.

  In a step 61, the module tz is calculated. In a step 62, the method then selects the coefficient Cn (k) corresponding to the calculated module zz. In a step 63, the method performs the calculation Cn (k) xz which is carried out. by block 5; In a step 64, the method calculates the difference between the composite signal 15 received and the signal at the output of block 5; In a step 65, the method performs the correlation calculation between the signal z and the error signal E: E xz * In a step 66, the coefficient Cn + 1 (k) is updated according to the formula Cn + 1 (k) = Cn (k) + t E z * Thanks to the adaptive mechanism according to the present invention, it is possible to operate an automatic adjustment of the parameters of the coefficients Cn (k) calculated to minimize the correlation between the two signals E and z , which ensures the adaptation of the mechanism even when the internal characteristics of the transponders change over time.

  In a preferred embodiment, the mechanism for adjusting the values of the parameters simultaneously performs a correction on two or three consecutive parameters Cn (k) and Cn (k +/- 1), with possibly different weightings within the table of correspondence implemented in block 5.

  ST - 03-GR2-081 - 14 In another embodiment, instead of calculating the module Iz], the value x2 + y2 is calculated directly and it is this value which is directly used to select the appropriate coefficient Cn ( k). This avoids extracting the square root which reduces the processing time.

ST - 03-GR2-081 - 15

Claims (11)

claims   1. Method for receiving a composite signal transmitted via a non-linear transmission channel according to a modulation scheme using a given constellation comprising a first signal UL and a second signal LL, characterized in that it comprises the steps following: - demodulation and decoding of said first UL signal by means of a first demodulation and decoding chain (2, 3) so as to regenerate said 1o first UL information; - recoding (4) from said first regenerated UL information a set of symbols z = x + j y representative of said constellation used in transmission and formatting to reconstruct a continuous time waveform of said reconstituted symbols; - applying to said continuous time waveform a self-adaptive non-linearity function (5); - subtract from said composite signal the result of said non-linearity function to generate a result E; - demodulation and decoding of said result E by means of a second demodulation and decoding chain (8, 9) so as to regenerate said second information LL.   2. Method according to claim 1 characterized in that said non-linearity function is performed by means of a correspondence table comprising p 25 coefficients Cn (k) with k = I to p and updated according to an adaptive process based on a correlation calculation between the result E of said subtraction and said continuous time waveform.   3. Method according to claim 1 or 2 characterized in that the application of said coefficients C, (k) of the non-linear function is carried out by means of a double Cartesian-polar-Cartesian transformation comprising the following steps: - transformation into coordinates polar of said input signal to determine the modulus and phase of said input signal; ST - 03-GR2-081 - 16 - apply the amplitude response F (p) and the phase response 0 (p) corresponding to the coefficient Cn (k) to be used; - polar / Cartesian conversion of the result.   4. Method according to claim 2 or 3 characterized in that said p coefficients of the correspondence table are distributed more densely around the saturation point of the transmission channel.   5. Method according to claim 2, 3 or 4 characterized in that the current coefficient o used for said non-linearity function is updated according to the following mechanism: Cn + l (k) = Cn (k) + pt E z * With p being a time constant, E being the result of said subtraction and z * being the conjugate complex of said symbol z = x + j y.   6. Method according to claim 2 characterized in that the coefficient Cn (k) used in said non-linearity function is directly determined from the calculation of x + y 7. Method according to claim 4 characterized in that two consecutive coefficients in said correspondence table are simultaneously updated for each correlation calculation between the result E of said subtraction and said continuous time waveform of the complexes z reconstituted.   8. Method according to one of the preceding claims, characterized in that it is applied to the transmission by satellite of a composite signal comprising a first primary information UL which can be received and processed by a large number of television receivers and secondary information LL which can be received and processed by a sub-category of receivers.   ST - 03-GR2-081 - 17 9. Receiver for a digital communication system intended to receive and process a composite signal transmitted via a non-linear transmission channel according to a modulation scheme using a given constellation comprising a first signal UL and a second signal LL, characterized in that that it comprises: - a first demodulation and decoding chain (2, 3) for demodulating and decoding said first UL signal so as to regenerate said first UL information; - a recoding and shaping block (4) for regenerating a continuous time waveform of said reconstituted symbols; o0 - a self-adaptive non-linearity function (5) applied to said continuous time function; - means for subtracting the result of said non-linearity function from said composite signal in order to generate a result E; - a second demodulation and decoding chain (8, 9) for demodulating and decoding said result E so as to regenerate said second information LL.   10. Receiver according to claim 9 characterized in that said non-linearity function is performed by means of a correspondence table comprising p 20 coefficients Cn (k) with k = 1 to p and updated according to an adaptive process based on a correlation calculation between the result E of said subtraction and said continuous time waveform.   11. Receiver according to claim 9 or 10 characterized in that the current coefficient 25 used for said non-linearity function is updated according to the following mechanism: Cn + 1 (k) = Cn (k) + p E z * With t being a time constant, E being the result of said subtraction and z * being the conjugate complex of said symbol z = x + j y. ST - 03-GR2-081